Researchers from the Massachusetts Institute of Technology have developed a new algorithm developed to Fourier transformations to apply to digital data. This would make data better and faster to compress.
The researchers presented their algorithm during the SODA, or the Symposium on Discrete Algorithms, which in recent days was held by the Association for Computing Machinery. The MIT staff developed a so-called fast Fourier transform algorithm that speedups of a factor of ten compared to the current fft-algorithms. Especially the compression of images would be of the new algorithm can take advantage of.
Fourier transformations are used in the conversion of analog signals to digital data and can be applied in signal processing, but also play a role in the compression of images. A complex signal with fft will be reduced to a few simple digitally to represent components. An analog sample is divided into small pieces, then the pieces will be analyzed and converted to a digital signal.
The MIT researchers developed first of all a better filter to the analog signal without signal loss in the particle splitting. Then the remaining frequencies are analyzed and weighted, whereby only the frequencies with a large contribution to the signal are kept. The algorithm analyzes the frequencies to do this, not only in strength but also in time, to fluctuations in measurements. This roads of the frequencies and ignoring unimportant or informatiearme frequencies, the fft of the MIT-team is suitable for compression.
Build a complex signal from simple components